Knowledge that Transforms

Mergers are associated with large and persistent earnings declines for incumbent employees in target firms. Linking employer‐employee administrative data with information on merger activity in Brazil, I find the negative effects concentrate on employees who exit target firms and reflect displacement in the short run and wage declines in the long run. Low‐skilled, managerial, and older employees fare worse. Overall, I conclude that mergers are followed by substantial reallocation costs reflecting losses of firm‐specific wage premiums, matching inefficiencies, and industry‐specific human capital depreciation, with employees transitioning to lower paying firms considered to be of lower productivity and employment value.

We study how public disclosure of politicians’ outside income affects their behavior. We exploit a disclosure reform targeting German federal MPs and tax-return data in a difference-indifferences setup using unaffected state MPs as controls. MPs increase their outside income by 24%, driven by likely right-leaning MPs. A representative survey experiment uncovers that right-leaning voters interpret outside income as a signal of competence and hard work, while left-leaning voters associate it with weaker representation. Consistent with this, we show that newspapers cover right-leaning MPs’ outside activities more favorably. Our findings suggest that politicians strategically use public disclosure as a signaling tool.

Developing countries face high pharmaceutical prices despite the presence of multiple generic producers. One key reason is excessive spending on sales and marketing by drug firms. This paper examines how centralized drug procurement can curb such marketing efforts and reduce prices, using a policy experiment in China where certain generics are procured through centralized auctions in pilot cities. Winning firms gain large market shares directly without high marketing costs. We find that the centralized auction significantly reduces both sales costs and prices, with winning firms experiencing notable decreases in marketing-related labor demand.

We study the unintended effects of Fair Access to Insurance Requirements (FAIR) plans developed by 26 states in the 1960s to address insurance redlining in urban neighborhoods. FAIR plans’ problematic features included prohibitions on considering environmental hazards in underwriting, mandatory insurer participation in pools that diluted underwriting incentives, and payouts exceeding market values in declining areas. Using a triple-difference design comparing pre/post-FAIR periods, neighborhoods with/without likely FAIR access, and participating/non-participating states, we find that FAIR inadvertently led to significant housing disinvestment and accelerated declines in neighborhood population, with simultaneous increases in the Black population share.

We study how employer concentration affects job design, focusing on schedule flexibility as a non-wage job attribute. Using the full text of 46 million UK job vacancies from 2014–2019, we construct new measures of flexible work arrangements using supervised machine learning and large-language-model—assisted annotations. Adopting an instrumental variable strategy using a Bartik-style prediction of the change in employer concentration based on nationwide hiring trends, we show that higher concentration reduces the value of worker compensation partly through shifts toward “risky” flexibility that exposes workers to earnings volatility.

This paper reevaluates the longstanding debate on capital-labor substitution by examining the role of external inputs: raw materials, intermediate goods and services, imports, offshoring. Both a meta-analysis (analyzing existing estimates of substitution) and direct estimation (using U.S. data for 1963-2016) indicate that external inputs disproportionately displace labor. These findings imply (1) that the capital-labor ratio responds 40-80% more strongly to the price of labor than to the price of capital, (2) that value added cannot be modeled separably from gross output, and (3) that historical disagreements regarding substitution can be recast as an omitted variable bias involving external inputs.

A recent literature studies fiscal policy with realistic intertemporal marginal propensities to consume and the absence of Ricardian equivalence. This literature shows that fiscal policy boosts consumption and is effective: (i) Cumulative fiscal multipliers are above one, and (ii) fiscal policy is fully self-financing. I show this does not extend to small open economies by providing conditions where (i) the cumulative fiscal multiplier is exactly one, and (ii) fiscal deficits are not fully self-financing. This is because the initial debt-fueled rise in consumption is offset by a drop to repay foreign debt. Finally, I provide conditions to overturn this result.

We identify the distributional and quantile marginal treatment effect functions when the outcome is right-censored. Our method requires a conditionally exogenous instrument and random censoring. We propose asymptotically consistent semi-parametric estimators and valid inferential procedures for the target functions . To illustrate, we evaluate the effect of alternative sentences (fines and community service vs. no punishment) on recidivism in Brazil. Our results highlight substantial treatment effect heterogeneity: we find that people whom most judges would punish take longer to recidivate, while people who would be punished only by strict judges recidivate at an earlier date than if they were not punished.

This paper studies how to design experiments when the experimental units are large aggregate entities (e.g., markets), and only one or a small number of units can be treated. We propose a class of non-randomized experimental designs using the synthetic control method, which jointly selects treated units and untreated units to serve as a comparison group. We analyze the properties of estimators under these synthetic control designs and develop new inferential techniques. We show that, in experiments with aggregate units, synthetic control designs can substantially reduce estimation bias relative to randomized experiments.